Melanopsin-mediated pupillary responses in bipolar disorder—a cross-sectional pupillometric investigation

In bipolar disorder (BD), a multitude of external factors can interfere with the course of disease and provoke the onset of an affective episode, i.e. depression or mania (Young and Dulcis 2015). Contributing factors include disturbances in biological rhythms caused by e.g. sleep deprivation or change of season. As such, there is evidence for a distinct seasonal pattern in health care utilization in BD with a peak rate of admissions for mania in spring/summer and for depression in winter (Geoffroy et al. 2014) just as patients with BD report increased fluctuations in mood and behavior in relation to change of seasons (Geoffroy et al. 2014). The main driver of this seasonal variation is presumably the change in daylight (Geoffroy et al. 2014; Aguglia et al. 2017). In line with this, chronotherapeutic modalities such as light and dark therapy can be used adjunctively in the treatment of depression and mania, respectively (Gottlieb et al. 2019; van Houl et al. 2020). A hypothesis of so-called retinal super-sensitivity to light has been suggested in BD (Lewy et al. 1985, 1981). Early support for this hypothesis included the findings of increased melatonin suppression in response to light in individuals with BD as well as in offspring of patients with BD (Lewy et al. 1985, 1981; Nurnberger et al. 1988). Other studies could not corroborate these findings (Lam et al. 1990; Whalley et al. 1991; Ritter et al. 2019). Aberrant melatonin suppression in patients and offspring could indicate that aberrant light sensitivity constitutes an inherited trait biomarker. Such endophenotypes can potentially inform on underlying biological vulnerability or predict illness emergence in un-affected carriers of the trait. State-dependent biomarkers reflect on the phenotypical presentation of current illness states and can potentially aid in diagnostic and staging processes.

As light reaches the retina, specialized neurons—rods, cones and intrinsically photosensitive retinal ganglion cells (ipRGCs)—perceive the signal (Hattar et al. 2002). The ipRGCs play essential roles in the transmission of light for non-visual purposes such as the pupillary light reflex, suppression of melatonin and regulation of diurnal rhythm, alertness and mood (Lazzerini Ospri et al. 2017; Fernandez et al. 2018). The ipRGCs are predominantly sensitive to high-intensity short wavelength—blue—light (Hattar et al. 2002; Park et al. 2011). IpRGC-activation by blue light elicits the so-called post-illumination pupillary response (PIPR),i.e. a pupillary constriction that persists past the termination of a light stimulus. The PIPR can be swiftly and non-invasively assessed with chromatic pupillometry (Park et al. 2011), i.e. recording and quantification of the pupillary responses to monochromatic light stimuli (Park et al. 2011). Pupillometric evaluation of the ipRGC-mediated responses may provide a potential alternative to the invasive, time consuming, and expensive assessment of circadian system light sensitivity by melatonin suppression experiments. The disadvantages of the method include numerous sources of intra- and inter-individual variation, such as circadian and seasonal timing, iris color, medication, etc. (Kelbsch et al. 2019).

There is some evidence that the PIPR is attenuated in persons with unipolar depression (Roecklein et al. 2013; Berman et al. 2018), but it has not been investigated in persons with BD. In persons without affective disorder, a more pronounced PIPR was associated with higher levels of self-reported lifetime hypomanic/manic-like experiences (Bullock et al. 2019). In this first pupillometric investigation in BD, we assessed trait and state biomarker characteristics of the ipRGC-mediated pupillary response in patients with newly-diagnosed BD and we explored potential associations with illness variables.

In a cross-sectional design, we investigated the pupillary responses in patients with newly diagnosed BD, their unaffected first-degree relatives and a healthy control group between September 2018 and November 2019. All participants attend the research facility for regular follow-up visits as a part of the Bipolar Illness Onset (BIO)-study (Kessing et al. 2017). BIO-participants who had no prior or current eye disorder, eye trauma, eye surgery or family history of glaucoma were invited to participate in the pupillometric study.

The study was approved by the Committee on Health Research Ethics of the Capital region of Denmark (H-7-2014-007) and complied with the Declaration of Helsinki. Participants all provided written informed consent.

From this first pupillometric assessment of patients with BD, we report indications that the ipRGC-mediated pupillary responses are altered along with the duration of illness and potentially across different illness states. However, we do not see any indications of aberrant ipRGC-responses being an inherited trait phenomenon in BD.

Although the retinal super-sensitivity hypothesis initially received support from a few studies reporting increased melatonin suppression in patients with BD and in unaffected adolescents with a high-risk profile for BD (Lewy et al. 1985, 1981; Nurnberger et al. 1988; Nathan et al. 1999), later studies challenged the hypothesis (Lam et al. 1990; Whalley et al. 1991; Ritter et al. 2019). Increased melatonin suppression was not found in remitted, drug-free patients (Ritter et al. 2019) nor in a recent well-controlled study using blue light for specific activation of the ipRGC pathway in euthymic patients with BD-I (Ritter et al. 2019). The latter study further reported that the subjective alerting response to blue light was attenuated in patients with BD compared with controls. This is much in line with the group of remitted patients in our study who displayed attenuated ipRGC-responses compared with controls. This finding was, however, based on only 9 remitted patients why it should be interpreted with caution. However, from the larger BIO-cohort, our group has shown, that the BD-group has disordered sleep compared with unaffected relatives and healthy controls even in remitted phases (la Cour Karottki et al. 2020). This inter-episode sleep disturbances may be associated with aberrant ipRGC-transmission. In healthy adolescents and young adults, a more pronounced PIPR is associated with a propensity for delayed sleep timing (Meijden et al. 2016).

We were not able to identify differences in the PIPR across different affective states or associations with symptom severity. This may relate to either sample size (type II error) or the mild-to-moderate symptom load in the present population. The median scores do indicate a trend toward a more pronounced PIPR in depressive and especially in hypomanic/mixed state compared with remission. The prior pupillometric studies in unipolar depression have not assessed the associations between the PIPR and symptom severity (Roecklein et al. 2013; Berman et al. 2018; Laurenzo et al. 2016; Feigl et al. 2018). Moreover, since the studies have used different instruments to assess depression severity, the populations are not directly comparable. The PIPR was similar in 8 young persons with mild-to-moderate depression and 13 healthy controls (Feigl et al. 2018). In another study of 19 persons with unipolar depression of whom the majority were in remission, there was a trend toward a 7% reduction of the PIPR, p = 0.07 (Laurenzo et al. 2016). In the largest study to date, 19 in- and outpatients with depression and 20 patients with SAD were assessed with pupillometry during a depressive episode in winter and at follow-up 6 months later (Berman et al. 2018). In both groups with depression, the PIPR was reduced compared with healthy controls. The participants were recruited from a hospital setting which could indicate a more severe illness than in the other studies although their depression scores categorized as mild-to-moderate. In the first pupillometric study in affective disorder performed in persons with SAD, the PIPR was attenuated compared with controls (Roecklein et al. 2013). In SAD, the reduction in ipRGC-responsivity has been suggested as a causal factor, but the association between symptom severity and PIPR remains unresolved in unipolar depression. State marker potential of the PIPR in BD should be evaluated in longitudinal designs of patients across different affective states and preferably with higher symptom severity.

We find a positive association between the PIPR_late and illness duration. Due to strong collinearity between age and illness duration, this association was not significant when adjusting for sex and age. As the PIPR is known to be stable across increasing age (Kankipati et al. 2010; Daneault et al. 2016) and since our data do not indicate an association between age and the PIPR_late in either the UR or HC groups, we cautiously suggest that the increase in PIPR relates more to illness duration or progression than to higher age. This would be in line with the concept of BD as a progressive disorder presented elsewhere (Kessing and Andersen 2017). A recent OCT study of inpatients with BD did indeed identify a thinning of the retinal nerve fiber layer in BD compared with healthy controls (Khalil et al. 2017). The aforementioned pupillometric studies in unipolar depression have not reported on number of prior episodes or illness duration, hence the association between illness duration and the PIPR constitutes a novel finding. If ipRGC-responsivity is indeed altered in more advanced or severe BD, chronotherapeutic modalities may be more effective in these groups. Blue light-blocking eyewear has been evaluated as an add-on treatment for severe mania in a randomized design in hospitalized patients with a mean illness duration of 20 years (Henriksen et al. 2016). In this population, there was a substantial and rapid reduction in YMRS scores with use of blue-blocking glasses eyewear compared with placebo glasses.

There is a number of limitations to consider in relation to our study. First and foremost, the sample size is modest and do not allow for adjustment for all potential confounding factors. We included only 3 patients with current manic symptoms, why we cannot evaluate any association between hypomania/mania and the PIPR. Importantly, we are also not able to decipher the potential effects of psychotropic medication, since all patients received medication and the majority received combination treatment. By their effects on neurotransmitter systems, psychotropic medications may potentially modify the pupillary motility and/or the sensitivity of the ipRGC-system. A recent study in unmedicated persons found that melatonin suppression was reduced in persons with current but not remitted depression (McGlashan ett al. 2018). No studies have specifically investigated the effects of psychopharmacologic agents on the blue light induced PIPR, but selective serotonin reuptake inhibitors (SSRI) and some antipsychotic agents have been shown to alter the pupil diameter (Koller et al. 2018; Bitsios et al. 1999). An acute dose of the antidepressant citalopram was shown to increase melatonin suppression without affecting the pupillary diameter in a double-blind, placebo-controlled crossover study (McGlashan et al. 2018). No modulatory effects on the PIPR has been reported from post-hoc analyses of antidepressant or anti-manic medications in the referenced pupillometric studies (Roecklein et al. 2013; Berman et al. 2018; Laurenzo et al. 2016; Feigl et al. 2018). Conclusively, medication effects on the pupillary responses or on the circadian system cannot be discarded, although there is no current evidence for drug-related effects on the ipRGC-related pupillary responses, but some evidence for alterations of the resting-state pupillary diameter. We report the PIPR_late as percentage pupil constriction relative to baseline in normalized values and not crude numbers, why differences in the resting state pupillary diameters are considered. Another limitation is that we were not able to perform full adjustments for potential effects of photoperiod and time of day due to the sample size. An association between the PIPR and photoperiod has been reported, i.e. a winter-to-summer difference limited to 1–2% (Berman et al. 2018; Laurenzo et al. 2016; Münch et al. 2017) that was primarily related to direct light exposure and not as much to photoperiod (Laurenzo et al. 2016). Our assessments were performed at random dates throughout the year, excluding the winter months, but we made no systematic attempts to match the groups for season or assess light exposure. The main outcome of our study, the comparison of the PIPR_late in the 3 groups was adjusted for seasonal but not for diurnal effects. The PIPR undergoes diurnal regulation with an increase in the early morning, a midday plateau and an evening decrease (Münch et al. 2012; Zele et al. 2011). To avoid these morning and evening extremes, we limited the time of assessment to normal office hours, where the PIPR is supposed to display a stable plateau-phase.

Strengths of our study include a thorough diagnostic evaluation to ensure the diagnosis of BD and absence of psychopathology in the two control groups. The inclusion of patients with early BD and their unaffected relatives allows us to specifically assess potential endophenotype features of the ipRGC-responses in BD. We assessed visual acuity and retinal structure to ensure that any pupillometric variation was not related to ocular pathology. Moreover, the implementation of pupillometry in affective disorder research has the potential to add to our understanding of the intertwining of the ipRGC-system and affective regulation. Retinal sensitivity measured during nighttime darkness in melatonin suppression designs may not correspond directly with daytime light sensitivity or ipRGC-function. In glaucoma, where ipRGC degeneration causes attenuations of the PIPR, the correlation between melatonin suppression and the PIPR is rather weak (Munch et al. 2015).

In this first investigation of the ipRGC-related pupillary responses in BD, we find a reduced ipRGC-response in remitted patients compared with controls, but no signs that such aberrant ipRGC-signaling constitutes an inherited trait feature in BD. The positive association between the ipRGC-responses and illness duration could suggest an alteration of the ipRGC-system over the course of disease. This new hypothesis needs corroboration in longitudinal designs. State marker characteristics of the ipRGC-mediated responses in BD should be tested in larger samples in longitudinal designs and include patients with a higher load of affective symptoms.